Retrieval-augmented generation (RAG)
By the Pickgap team · Updated Oct 15, 2026 · 1 min read
Retrieval-augmented generation (RAG) is a method where an AI system first retrieves relevant documents, such as web pages from a search, then writes its answer using them. AI search assistants work this way. For a business, it means AI answers are shaped by which pages get retrieved, which is why being on well-ranked, relevant pages matters.
The two steps
- Retrieve. The system searches a source, like the web, a search index or a database, for documents relevant to the question.
- Generate. It passes those documents to the language model, which writes an answer based on them, often with citations.
The quality of the answer depends heavily on step 1: the model can only use what was retrieved.
RAG in AI search
ChatGPT search, Perplexity, Gemini, and Google's AI Overviews and AI Mode all use retrieval in some form. The details differ: which search index, how many pages, whether the question is split into sub-searches (see query fan-out).
What it means for local businesses
To be in an AI answer, information about you needs to be in the retrieved pages. That means:
- Being listed on the directories and sites that rank for questions about your trade and city.
- Having pages on your own site that match specific questions.
- Making sure crawlers can read your site.